- Duke University

Download Report

Transcript - Duke University

For Pacific EBM Practitioners
Jason Roberts and Eric Treml
27-Aug-2008
Duke Marine Geospatial Ecology Lab
Duke Main Campus
Durham, North Carolina
Washington, D.C.
Jason Roberts
Duke Marine Lab
Beaufort, North Carolina
Caroline Good
Connie Kot
Elliott Hazen
Lab Director:
Dr. Patrick N. Halpin
Sarasota, Florida
Daniel Dunn
Staff and Students:
Ben Best
Andre Boustany
Andrew Dimatteo
Ben Donnally
Ari Friedlaender
Ei Fujioka
Erin Labrecque
Rob Schick
University of Queensland
Brisbane, Australia
Eric Treml
MGEL OVERVIEW
The Marine Geospatial Ecology Lab
(MGEL) applies geospatial
technologies to issues in marine
ecology, resource management and
ocean conservation.
Marine Ecology
Marine Geospatial
Ecology Lab
Informatics
Biogeography
The conservation and management of
natural resources in the marine realm
all occur in a spatial context.
We synthesize approaches from
Spatial Ecology, Biogeography, and
Geostatistics as well as geospatial
technologies, including geographical
information systems (GIS), global
positioning systems (GPS) and remote
sensing.
RESEARCH PROJECTS
Biogeographic Information Systems
The OBIS-SEAMAP
archives geo-referenced
data on marine mammals,
seabirds and sea turtles
globally.
The ESRI Marine Data
Model facilitates the
dynamic representation of
marine features in time
and space for GIS users.
Behavioral Ecology
Analysis of Marine
Animal Movement
provides new insights into
pelagic animal behavior
and management.
The Southern Ocean
GLOBEC Program
investigates the ecology
and behavior of large
whales in the Antarctic.
In collaboration with
Stanford University, the
Tag-A-Giant tracking
project studies the
movement and ecology of
bluefin tuna..
Ecosystem-Based Management
Project GloBAL assesses
global fishing effort and
fisheries bycatch.
The Packard-funded
Marine-EBM Tool
Innovaton Fund supports
the development of marine
EBM software tools.
Connectivity and Reserve Design
In collaboration with the
University of Queensland,
Coral Reef Connectivity
is being quantified with
hydrodynamic models and
graph theory.
Marine Eco-Regional
Plans with The Nature
Conservancy identify
conservation priorities in
the southeast USA.
Habitat Modeling
In collaboration with
NOAA, the SERDPfunded marine mammal
habitat modeling project
forecasts whale habitat for
the US Navy.
The Marine Geospatial
Ecology Tools provide
habitat, benthic, and
connectivity modeling
tools within scientific
workflows.
Talk outline
 Overview of Marine Geospatial Ecology Tools (MGET)
 Quick tour of MGET via an example user scenario
 Connectivity tools
 Questions
What is MGET?
 A collection of geoprocessing tools for marine ecology
 Oceanographic data management and analysis
 Habitat modeling, connectivity modeling, statistics
 Highly modular; designed to be used in many scenarios
 Emphasis on batch processing and interoperability
 Free, open source
 Written in Python, R, MATLAB, and C++
 Designed mainly for intermediate-skill ArcGIS users
 Minimum requirements: Win XP, ArcGIS 9.1, Python 2.4
 ArcGIS and Windows are only non-free requirements
MGET interface in ArcGIS
The MGET toolbox appears in the ArcToolbox window
MGET interface in ArcGIS
 Drill into the toolbox to find the tools
 Double-click tools to execute directly, or drag to
geoprocessing models to create a workflow
Interoperability
MGET “tools” are really just Python functions with input
and output parameters:
def DoSomething(input1, input2, output1)
Python programmers can call MGET functions directly.
To facilitate interoperability, MGET exposes these
functions as COM Automation objects and ArcGIS tools.
COM-capable
program:
C / C++ / C#,
Visual Basic,
R, MATLAB,
Java, etc.
MGET COM
Automation
DoSomething
class
ArcGIS geoprocessing tool
Example scenario
 Practitioner has spatially-explicit observations of a
species and wants to investigate:
 Are there spatial and temporal patterns?
 Correlations with environmental conditions?
 Correlations with occurrences of other species?
 Can we predict its occurrence and thereby
improve our management of it?

By designing MPAs, for example
Typical observation data
Fishery catch and
bycatch records
Surveys
IATTC
Olive Ridley
Encounters
1990-2005
Argos satellite
tracks
Figure courtesy of Scott Eckert
Analytic approach
1. Review scientific literature or consult an expert to
determine the environmental parameters that
might affect the distribution of your species.
2. If possible, measure those parameters when you
observe the presence or absence of the species.
3. For parameters you can’t measure at observation
time, use software tools to obtain them from
remotely-sensed oceanographic data.
4. Build and analyze statistical models that try to
predict the species’ distribution.
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
MGET includes tools that
assist with all of these steps
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
Species observations
 Skipping the details of this step to save time
 Ultimately you must produce a point shapefile or
feature class that shows locations where the species
was present and where it was absent Species presence field:
1 = present, 0 = absent
Date field
records
date of
observation
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Download
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
Options for obtaining data
1. Download files from data providers using FTP
 Nearly all data products are available with FTP
 Powerful, free downloaders exist (e.g. SmartFTP)
 But must often convert files to ArcGIS-compatible formats
2. Download using MGET or other tool (e.g. NOAA EDC)
 The tool hides details of download, using FTP, OPeNDAP
or other protocols, and writes ArcGIS-compatible formats
 Not many such tools exist
3. Order files on CD-ROM or DVD-ROM
 Use this if your Internet connection is slow
Example MGET tool for downloading data:
Download Aviso SSH Dataset to ArcGIS Rasters
Example SSH and currents data with turtle track
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
Preparing oceanography for use
 Most oceanographic datasets are not immediately
usable by ArcGIS
 Common preprocessing steps include:
 Converting to an ArcGIS-supported format
 Projecting to a desired projection
 Clipping to region of interest
 Performing basic calculations (via map algebra)

E.g. converting integers given by the original data
provider to floats that represent the real values
 Building pyramids
A very popular MGET tool:
Convert HDF SDS to ArcGIS Raster
Sea surface temperature
NOAA CoastWatch AVHRR
GOES 10/12
from PO.DAAC
NOAA NODC 4km AVHRR Pathfinder v5
Also: MODIS Aqua
and Terra, GOES 9
Sea surface chlorophyll density
SeaWiFS from the NASA GSFC OceanColor Group
Also: MODIS Aqua and combined MODIS/SeaWiFS
QuikSCAT ocean winds from PO.DAAC
28-Aug-2005
Katrina
Also: BYU
QuikSCAT
Sigma-0
(approximates
sea surface
rougness)
Global bathymetries
 ETOPO2
 GEBCO
 S2004
Map shows S2004
clipped to eastern
Pacific ocean
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
AVHRR Daytime SST
03-Jan-2005
Identifying Surface Temperature Fronts
Cayula-Cornillion Edge Detection Algorithm (1992)
Mexico
Step 1: Histogram analysis
Geoprocessing model
Frequency
Bimodal
Optimal
break
27.0 °C
Temperature
Example output
Step 2: Spatial cohesion test
Mexico
28.0 °C
Front
25.8 °C
Strong cohesion
 front present
~120 km
Weak cohesion
 no front
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
or behavior
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
Sampling
Sampling is the procedure of overlaying points over a map
and storing the map’s value as an attribute of each point.
Chlorophyll-a Density
Chl field filled with
values from the map
MGET has sampling tools
for various scenarios
Typical workflow
Import species
observations
into GIS
Analyze/model
species habitat
with statistics
Obtain
oceanographic
datasets
Explore maps
of oceano. and
observations
Sample
oceanographic
data
Prepare
oceanographic
data for use
Create derived
oceanographic
datasets
MGET statistics tools
 Lots of tools, many
more planned
 Built from Ben
Best’s ArcRStats /
HabMod projects
 Tools require the R
statistics program
to be installed on
your computer
Exploratory analysis
Density Histogram tool
2.0e-06
Turtle present
Density
Density
1.5e-06
Turtle absent
1.0e-06
5.0e-07
0.0e+00
0
500000
1000000
data$ArribadaBeachDist
Nesting
season
Distance
to nesting
beach (m)
Scatterplot Matrix tool
Fitting statistical models
ROC plots
Term plots
Predicting habitat maps from the model
Input #3:
Rasters for
predictor
variables
Input #1:
The fitted model
Predict GAM tool
Binary habitat (cutoff = 0.025)
Input #2:
Cutoff value
Predicted species presence
Bayesian probability that
predicted presence ≥ 0.025
Marine Connectivity
via Larval dispersal
Many marine species
• Larval dispersal stage
• Sessile or sedentary adults
• Space/habitat limited
• Ocean currents influence transport
• Populations connected via dispersal
Modeling Connectivity
Study site
Simulating the release of larvae from
450 reefs in the Tropical Pacific
- coral reefs
Modeling Connectivity
Quantifying arrival probability
Modeling larval dispersal via ocean currents
using Eulerian advection/diffusion
Modeling Connectivity
Graph Structure
Summarizing reef-to-reef
dispersal using directed graphs
Graph Analysis
Example: Potential Sources and Sinks
Applying graph theory to better understand connectivity and inform conservation
MGET coral reef connectivity tools
Coral reef ID and % cover maps
Ocean currents data
Tool downloads data for the
region and dates you specify
Edge list feature class representing
dispersal network
Larval density time series rasters
Thanks!
Download MGET:
http://code.env.duke.edu/projects/mget
Jason Roberts
[email protected]
Eric Treml
[email protected]